Anthony Harrington looks at the debate between minimum variance strategists and risk-efficiency strategists. Is one solution merely a staging post on route to the other?
Before the financial crisis, when everyone was dancing as long as the music kept playing, the current appetite for and debates around low-volatility equity strategies would have seemed very odd indeed. But five years is a long time and a 40% drawdown represents big losses.
Despite the discussion and the inflows, however, it remains true that pension funds move slowly when they are offered new products, and even more slowly when they are offered multiple competing products based on similar themes but in rather different flavours.
If we look at low-volatility or minimum variance strategies as a departure from the traditional market cap-weighted index benchmark that investors know and (used to) love, it is clear that they have at least three major rivals among systematic quantitative strategies (leaving aside the wildly successful fundamental indexation products). There is the ‘maximum diversification’ approach pioneered by Yves Choueifaty, the CEO of TOBAM; the risk parity approaches espoused chiefly by US houses such as Bridgewater, AQR and PanAgora; and the ‘efficient indexation’ strategy developed by the French business school, EDHEC.
So is low volatility simply an early, naive arrival on the scene that will be upstaged by ‘more efficient’ portfolio construction strategies? The answer depends very much on who you speak to, and all of these solutions really need a mid-term, five to seven-year cycle to prove themselves – which probably means that this space is not going to get any less murky for some time. Today, investors might therefore be best advised to allocate a portion of their equity portfolios to non cap-weighted strategies, and then split this portion among some or all of the approaches.
According to Susanne Willumsen, portfolio manager for the Global Controlled Volatility fund at Lazard Asset Management, the answer to the question of whether low volatility will be upstaged, is, as one might expect, emphatically ‘no’. Low-volatility strategies – leavened, she emphasises, by fundamental insights – are the real deal. Lazard’s approach is highly scalable and Willumsen and her colleagues are setting out their stall to ramp the fund up as pension funds come around to their way of thinking.
“The plain fact is that institutional investors need equity as a key building block of their asset mix,” she says. “Without it, they cannot get to a fully funded position without massive support from the scheme sponsor. Low-volatility equities are a solution for those who cannot tolerate the volatility in asset prices, but who need the long-term capital appreciation that equity brings. The evidence – and our experience – shows that you can achieve equity market returns with a strong risk/return trade-off, through a portfolio comprising a fully diversified range of low-volatility stocks.”
When it comes to the question of which way is best to achieve that risk/return profile, Willumsen suggests that the main message to get across to pension fund trustees is the vital importance of the initial move away from the market cap-weighted benchmark. Once you decide to do that, there are all kinds of things investors can do to deliver a much more interesting risk/return.
“We are trying to end up with the lowest risk portfolio,” she says. “This is different from the maximum diversification approach [pursued by TOBAM], where you end up with a portfolio that is less than the index in terms of risk, but not much less.”
She also contrasts Lazard’s low-volatility strategy with another influential break from the cap-weighted benchmark – fundamental indexation and other value-biased approaches. Here, again, the risk premium associated with value and smaller companies adds volatility, she says, particularly relative to the larger-cap, more liquid, low-volatility positions that Lazard focuses on.
Erik Gosule at PanAgora argues that risk parity is the best way to go. The whole idea is to balance the various categories of risk across the fund so that no single, concentrated source of risk can have a meaningful negative impact on the performance of the fund as a whole.
“We are looking to create a risk-balanced portfolio,” he claims. “This will give you a similar outcome, in many circumstances, to Lazard’s low-volatility portfolio, but it is the result of trying to create a risk-efficient portfolio as opposed to trying to optimise an investment strategy built around low volatility.”
The PanAgora approach has more-or-less the same philosophical premise as that espoused by Choueifaty at TOBAM – namely that the best way of capturing the true equity risk premium is to construct a portfolio that has no bias to any one source of risk. If you balance out all risks then you are, by definition, favouring none; or, to put it another way, you are constructing a portfolio that takes no bets, explicit or implicit, on what elements are likely to outperform in future. It follows that PanAgora and TOBAM would both regard low-volatility funds as an attempt to bet on just another bias – towards low rather than high volatility.
“That’s fine if that bias leads to an upside,” reasons Gosule. “But it also has bubbles, and when these burst you get serious drawdowns.”
The technically challenging part about PanAgora’s approach is that you have to be in a position to identify the full spectrum of the risks that come from all possible exposures – from the oil price to interest rates to a rally in cyclical stocks. The list is potentially endless.
“You can’t necessarily focus on long-only value investing for years then suddenly say, ‘I think I’ll go for a risk parity approach now’,” says Gosule. “It is very challenging.”
For Choueifaty, it is too challenging. He thinks that claiming to have a handle on all possible sources of risk gets you into esoteric territory. He wishes good luck to those who reckon they can do it, but points out that it is far more difficult than his approach, which aims simply to create the most diversified portfolio available.
He asks us to imagine that there only two stocks, A and B, and that A has a volatility of 30%, while B has a volatility of 15%, and that there is a 60% correlation between them. The best ratio split between them will be one-third A combined with two-thirds B: this will gives the maximum-diversification portfolio. Similarly, if you have a universe of three stocks, A, B and C, where A and B have a correlation of 95% with one another and C has a correlation with A and B of just 5%, the maximum-diversification portfolio has 26% in stock A, 26% in B and 48% in C. But the thing to note in all of this is not only that at no point does this require anyone to make any speculative assumptions about the kinds of returns that you will get from A, B or C; it also requires no assumptions about the kinds of risk emanating from the stocks beyond their volatilities and correlations.
“When you have built the most diversified portfolio available to you, as a quantitative manager with a maximum diversification strategy, your job is done,” says Choueifaty. “You have absolutely no way of extrapolating from that portfolio to make any kind of prediction about the potential success or failure of A, B, and C as companies and equally, no way of telling what kinds of returns investing in their stock is likely to deliver. What you do have is a great deal of statistical evidence that shows that taken over a medium to long-term period, a maximum-diversified portfolio delivers greater returns than portfolios with any built-in bias.”
EDHEC’s ‘Risk-Efficient index’ products, developed with FTSE, have a very similar starting point to Choueifaty’s maximum diversification. Its aim is to construct portfolios for a maximum Sharpe ratio – identical to traditional portfolio optimisation approaches with the one exception that the usual expected return estimates are replaced by more predictable expected risk estimates.
“We know that the volatility, when correctly calculated for a period, has a certain forecasting value for the volatility of the following period,” explains head of applied research Felix Goltz. “Ultimately, the efficient maximum Sharpe ratio can be compared to the minimum-volatility approach, which is also a way of constructing an efficient portfolio without worrying about estimating the stocks’ returns. These two approaches are based on an estimation of the variance-covariance matrix, which is the main ingredient of any portfolio optimisation.”
Where EDHEC’s approach differs from a minimum volatility approach is that it affirms the traditional finance theory link between risk and reward and acknowledges that the riskiest stocks are the most profitable over the long term. As a result it avoids a concentration on low-volatility stocks and, in Goltz’s view, enables the diversification budget constituted by the correlations between stocks to be used better.
And it is this point about the diversification budget that gets to the heart of the debate about the low-volatility versus risk-efficient approaches to breaking away from the cap-weighted benchmark. Once correlation is taken into account, higher-volatility stocks can be seen as making a contribution to a lower-volatility portfolio. Indeed, many low-volatility portfolios implicitly accept this when they introduce variance-covariance optimisation precisely to prevent concentrations in ‘defensive’ sectors.
Those who argue for risk-efficient portfolios would ask, if the main benefit that investors get from low-volatility strategies is really the result of better diversification, when measured against the cap-weighted benchmark, and if low-volatility strategists accept this in their own portfolio construction techniques, why not take the next logical step and focus on the goal of diversification rather than the goal of low volatility?
The low-volatility apologists respond that there is more to their strategy than a simple diversification benefit. And it has to be said that long periods of empirical historical evidence support their claims – even as those claims offend against the efficient-market underpinnings of traditional portfolio theory and the newest iterations of efficient-portfolio construction. The only sure thing is that one is most certainly a bet on the continued success of a particular risk attribute, while the others are the studied refusal to make any such bets. There may be a place in a broad portfolio for both concepts.